See Editorial on Page 1289
Identification of liver transplant candidates with hepatocellular carcinoma and a very low dropout risk: Implications for the current organ allocation policy
Article first published online: 27 NOV 2013
© 2013 American Association for the Study of Liver Diseases
Volume 19, Issue 12, pages 1343–1353, December 2013
How to Cite
Mehta, N., Dodge, J. L., Goel, A., Roberts, J. P., Hirose, R. and Yao, F. Y. (2013), Identification of liver transplant candidates with hepatocellular carcinoma and a very low dropout risk: Implications for the current organ allocation policy. Liver Transpl, 19: 1343–1353. doi: 10.1002/lt.23753
This work was presented at the annual meeting of the American Association for the Study of Liver Diseases in Boston, MA, on November 10, 2012, and it was supported by a grant from the National Institutes of Health to the Liver Center at the University of California San Francisco (PO1-DK26743).
- Issue published online: 27 NOV 2013
- Article first published online: 27 NOV 2013
- Accepted manuscript online: 23 SEP 2013 07:31AM EST
- Manuscript Accepted: 25 AUG 2013
- Manuscript Revised: 5 AUG 2013
- Manuscript Received: 26 MAY 2013
It has been shown that patients with hepatocellular carcinoma (HCC) meeting the United Network for Organ Sharing T2 (Milan) criteria have an advantage in comparison with patients without HCC under the current organ allocation system for liver transplantation (LT). We hypothesized that within the T2 HCC group, there is a subgroup with a low risk of wait-list dropout that should not receive the same listing priority. This study evaluated 398 consecutive patients with T2 HCC listed for LT with a Model for End-Stage Liver Disease exception from March 2005 to January 2011 at our center. Competing risk (CR) regression was used to determine predictors of dropout. The probabilities of dropout due to tumor progression or death without LT according to the CR analysis were 9.4% at 6 months and 19.6% at 12 months. The median time from listing to LT was 8.8 months, and the median time from listing to dropout or death without LT was 7.2 months. Significant predictors of dropout or death without LT according to a multivariate CR regression included 1 tumor of 3.1 to 5 cm (versus 1 tumor of 3 cm or less), 2 or 3 tumors, a lack of a complete response to the first locoregional therapy (LRT), and a high alpha-fetoprotein (AFP) level after the first LRT. A subgroup (19.9%) that met certain criteria (1 tumor of 2 to 3 cm, a complete response after the first LRT, and an AFP level ≤ 20 ng/mL after the first LRT) had 1- and 2-year probabilities of dropout of 1.3% and 1.6%, respectively, whereas the probabilities were 21.6% and 26.5% for all other patients (P = 0.004). In conclusion, a combination of tumor characteristics and a complete response to the first LRT define a subgroup of patients with a very low risk of wait-list dropout who do not require the same listing priority. Our results may have important implications for the organ allocation policy for HCC. Liver Transpl 19:1343-1353, 2013. © 2013 AASLD.
Model for End-Stage Liver Disease
percutaneous ethanol injection
Response Evaluation Criteria in Solid Tumors
United Network for Organ Sharing
The development and validation of the Model for End-Stage Liver Disease (MELD) score for predicting mortality in patients with end-stage liver disease resulted in the implementation of this continuous scoring system by the United Network for Organ Sharing (UNOS) for the prioritization of patients for liver transplantation (LT) in 2002.[2, 3] Patients with hepatocellular carcinoma (HCC), on the other hand, may have well-preserved liver function and low calculated MELD scores, but they are at risk for tumor progression and dropout from the LT waiting list, especially if the wait-list time exceeds 6 months. Consequently, attempts were made to give patients with HCC extra priority based on the expected risk of dropout from the waiting list due to tumor progression and to establish equitable dropout rates for HCC and non-HCC patients. This priority system for HCC has so far been applied only to those meeting the Milan criteria. In the UNOS staging classification, the Milan criteria are further divided into stages T1 (1 lesion of <2 cm) and T2 (1 lesion of 2-5 cm or 2-3 lesions, each ≤ 3 cm). It became evident that HCC patients were initially being given excessive priority because of overestimations of their risks of dropout from the waiting list. After several modifications of the HCC MELD algorithm (most recently in January 2005), patients with stage T2 HCC are now given a lower MELD exception score that starts at 22 points; this is equivalent to a 15% risk of mortality at 3 months. They are eligible for continued upgrades at 3-month intervals (equivalent to a 10% increase in mortality) as long as the tumors remain within the T2 criteria. Additionally, since 2004, patients with T1 HCC have no longer been eligible for priority listing, in large part because of the high rate of HCC misdiagnosis in these patients.
Even with each of these changes in the organ allocation policy giving reduced priority to patients with T2 HCC, data have continued to emerge suggesting that patients with HCC retain a significant advantage in comparison with patients without HCC on the LT waiting list.[8-12] This is a challenging problem, and the solution remains unclear. A confounding factor is the regional variation in this discrepancy. Using a different organ allocation scheme to give priority to HCC patients at a high risk for dropout[9, 13] may lead to the selection of patients with a more aggressive tumor biology for LT.[10, 14] Another important consideration is the increased use of locoregional therapy (LRT) for HCC as a bridge to LT, which may influence the risk of dropout from the waiting list. A previous report from our center showed wait-list dropout rates of 11% at 6 months and 57% at 12 months, but only 41% of patients received LRT. In contrast, a number of subsequent studies applying transarterial chemoembolization (TACE) or radiofrequency ablation (RFA) to all HCC patients have shown wait-list dropout rates of 0% to 5.8% over a mean wait-list time of 5.9 to 12.7 months,[15-17] but the sample sizes have been small. On the other hand, large studies using national databases to identify predictors of dropout from the waiting list have not been able to assess the effects of LRT.[9, 13, 18]
Our center is within region 5, which is amongst the longest average wait-list times for HCC and non-HCC patients. This has allowed us to evaluate the patterns and risks of dropout based on tumor characteristics.[4, 19] We hypothesized that the response to LRT is an important determinant of the risk of wait-list dropout and may help to identify a subset of patients with a very low rate of wait-list dropout who do not warrant a listing priority equal to that of other patients with T2 HCC. We, therefore, evaluated clinical and radiographic factors, including the response to LRT, as possible predictors of dropout in a large cohort of patients with T2 HCC who were listed for LT at our center with a MELD exception since the latest change in the organ allocation policy for HCC in 2005.
PATIENTS AND METHODS
Study Design and Patient Population
This was a retrospective cohort study of patients with T2 HCC who were 18 years old or older and were listed for LT with a MELD exception from March 2005 to January 2011 at our center. The start date was chosen to encompass the most recent HCC policy change, by which the initial MELD exception was down-graded to 22 points for T2 HCC. One hundred fourteen of the 512 patients who were initially evaluated were excluded from the study. These 114 patients included 72 patients who underwent tumor down-staging to meet T2 criteria under a down-staging protocol as previously described, 18 patients who underwent transplantation at another institution, 10 patients who were lost to follow-up within 1 month of transplant listing, 7 patients who were found on liver explant to have cholangiocarcinoma and not HCC, and 7 patients who underwent living donor LT. The final cohort consisted of the remaining 398 patients.
The collected variables included demographic data (age, sex, and race/ethnicity), the size and number of tumors at the time of listing, laboratory data at the time of listing [alpha-fetoprotein (AFP) level and MELD score], and liver-related factors [etiology of liver disease and Child-Pugh-Turcotte (CPT) score]. Since September 2004, decisions regarding the management of patients with HCC awaiting LT have been made at a weekly multidisciplinary liver tumor board attended by transplant hepatologists, transplant surgeons, oncologists, interventional radiologists, and radiologists with expertise in diagnostic abdominal imaging. The diagnosis of HCC was based on radiographic features characteristic of HCC, which included arterial phase enhancement and portal venous phase washout for lesions measuring at least 1 cm. Nodules not showing these characteristics were considered indeterminate for HCC, and they were followed by computed tomography or magnetic resonance imaging every 3 months so that they could be observed for interval growth and be re-reviewed by the tumor board. Percutaneous biopsy was performed for the purpose of diagnosing HCC only in select cases with atypical radiographic features. Nodules < 1 cm were not counted as HCC.
We applied LRT and used repetitive interventions if they were needed to induce the complete necrosis of all tumor nodules if possible before LT. The decision regarding HCC treatment modalities was based on a review of imaging studies at our weekly multidisciplinary tumor board. The choice of LRT was based on the degree of tumor hypervascularity, the tumor location, the response to prior LRT, and the hepatic and functional reserve of the patient. In accordance with our protocol, patients underwent contrast-enhanced computed tomography or magnetic resonance imaging 1 month after each LRT and at a minimum of once every 3 months after they were listed for LT. The tumor response to the first LRT, regardless of whether it occurred before or after LT listing, was assessed through the collection of AFP levels before and after LT within 1 month of the treatment date and on the basis of radiographic responses according to the modified Response Evaluation Criteria in Solid Tumors (RECIST).[22, 23]
For patients who underwent LT, the explant pathology was reviewed to determine the histological grade according to the modified Edmondson criteria [(1) well differentiated, (2) moderately differentiated, or (3) poorly differentiated], the tumor stage, and the presence of vascular invasion. Explant tumor staging in this study was based on the size and number of only viable tumors. This study was approved by the University of California, San Francisco Institutional Review Board.
The primary outcome was dropout from the transplant waiting list for any of the following reasons: death without LT, HCC tumor progression beyond T2 criteria, or being too sick or medically unsuitable to undergo LT. The time to dropout or LT was measured from the date of the initial HCC listing to wait-list removal or LT, respectively. For patients who were removed from the waiting list if they were no longer interested in undergoing LT or were noncompliant with our center's transplant policies, follow-up was censored at the time of delisting. The date of death was obtained from our transplant database and was confirmed with the Social Security Death Index.
Patient characteristics were summarized with medians and ranges for continuous variables and with proportions for categorical variables. Patients on the transplant waiting list were at risk for several competing outcomes, including LT, dropout from the waiting list for any reason, and death without LT, so the cumulative incidence of dropout was estimated while we accounted for competing risks (CRs). Univariate and multivariate analyses of predictors of wait-list dropout were performed via CR regression with the Fine and Gray method. Predictors of dropout with a univariate P value less than 0.1 were evaluated in the multivariate analysis, with the final model selected by backward elimination (P for removal > 0.05). The model identified a subpopulation of patients with a low risk of wait-list dropout. The cumulative incidence for the low-dropout-risk group versus all other patients was estimated from the baseline function of the multivariate CR model. The clinical characteristics of the 2 groups were compared with Pearson's chi-square and Wilcoxon tests as appropriate. Statistical analyses were performed with Stata/IC 11.1 (StataCorp, College Station, TX).
The baseline demographic and clinical characteristics of the 398 patients composing the study population are summarized in Table 1. The median age was 57 years, and 76.9% were men. Caucasians (42.7%) and Asians (31.4%) made up the majority of the study population. Hepatitis C virus was the most common etiology of liver disease (60.6%) and was followed by hepatitis B virus (24.4%). At the time of listing with MELD exception points based on HCC, the median actual MELD score was 11. The median CPT score was 7: 46.7% of the patients were classified as Child class A (CPT score = 5-6), 41.5% were classified as Child class B (CPT score = 7-9), and 11.8% were classified as Child class C (CPT score ≥ 10). The median AFP level was 13 ng/mL at the time of LT listing. More than half of the patients (57.8%) had an AFP level ≤ 20 ng/ml, and 6.5% had an AFP level greater than 1000 ng/mL. When we subcategorized patients within T2 criteria, 36.9% had a single HCC lesion of 2 to 3 cm, 35.2% had a single HCC lesion 3.1 to 5 cm, and 27.9% had multiple lesions. Percutaneous biopsy was performed to confirm the diagnosis of HCC before LRT in 33 patients (8.3%).
|Median age (years)||57 (range = 21–77)|
|Male sex [n (%)]||306 (76.9)|
|Race/ethnicity [n (%)]|
|African American||24 (6.0)|
|Liver disease etiology [n (%)]|
|Hepatitis C virus||241 (60.6)|
|Hepatitis B virus||97 (24.4)|
|Nonalcoholic fatty liver disease||21 (5.3)|
|Median MELD score at listing||11 (range = 6–27)|
|MELD score at listing [n (%)]|
|Child-Pugh class [n (%)]|
|Median AFP level at listing (ng/mL)a||13 (IQR = 5–81)|
|AFP level at listing [n (%)]a|
|≤20 ng/mL||229 (57.7)|
|21–100 ng/mL||80 (20.2)|
|101–300 ng/mL||38 (9.6)|
|301–500 ng/mL||15 (3.8)|
|501–1000 ng/mL||9 (2.3)|
|>1000 ng/mL||26 (6.5)|
|Tumor number and size at diagnosis [n (%)]|
|1 lesion, 2–3 cm||147 (36.9)|
|1 lesion, 3.1–5 cm||140 (35.2)|
|2 lesions||80 (20.1)|
|3 lesions||31 (7.8)|
Radiographic and AFP Responses to LRT
The vast majority of the patients (96.5%) underwent at least 1 LRT after the diagnosis of HCC was made. TACE was the most commonly used modality for a patient's first LRT (71.7%). A complete response on imaging 1 month after this initial treatment was seen in 60.1% of the patients according to the modified RECIST criteria, whereas a partial response (20.5%) or stable disease (17.2%) was seen in a majority of the remaining patients. Only 8 patients (2.2%) had progressive disease. The median AFP level decreased from 18.0 ng/mL before the first LRT to 9.5 ng/mL after the first LRT. Approximately two-thirds of the patients (66.1%) had an AFP level ≤ 20 ng/mL after they received their first LRT (Table 2).
|Number of LRTs received [n (%)]a|
|Type of first LRT received [n (%)]b|
|Response to first LRT according to modified RECIST criteria [n (%)]c|
|Complete response||220 (60.1)|
|Partial response||75 (20.5)|
|Stable disease||63 (17.2)|
|Progressive disease||8 (2.2)|
|Median AFP level before first LRT (ng/mL)d||18.0 (IQR = 6–102)|
|AFP level before first LRT [n (%)]d|
|≤20 ng/mL||186 (51.8)|
|21–100 ng/mL||82 (22.8)|
|101–300 ng/mL||31 (8.6)|
|301–500 ng/mL||17 (4.7)|
|501–1000 ng/mL||15 (4.2)|
|>1000 ng/mL||28 (7.8)|
|Median AFP level after first LRT (ng/mL)e||9.5 (IQR = 4.6–37)|
|AFP level after first LRT [n (%)]e|
|≤20 ng/mL||218 (66.1)|
|21–100 ng/mL||65 (19.7)|
|101–300 ng/mL||19 (5.8)|
|301–500 ng/mL||7 (2.1)|
|501–1000 ng/mL||9 (2.7)|
|>1000 ng/mL||12 (3.6)|
Outcomes on the Waiting List
Two hundred seventy-one of the 398 patients in the cohort (68.1%) underwent LT with a median waiting time of 8.8 months [interquartile range (IQR) = 5.9-12.8 months]. Sixty-six patients (16.6%) dropped out because of tumor progression, and 26 patients (6.5%) died while they were on the waiting list. The median time from listing to dropout due to tumor progression or death without LT was 7.2 months (IQR = 3.5-10.6 months). The cumulative probabilities of dropout due to tumor progression or death according to a CR analysis were 9.4% [95% confidence interval (CI) = 6.8%-12.5%] within 6 months and 19.6% (95% CI = 15.8%-23.7%) within 12 months. Twenty-seven patients were censored at the time of their wait-list removal. The reasons for wait-list removal included significant cardiopulmonary disease precluding LT (n = 6), noncompliance (n = 4), and a decision by the patient not to undergo LT (n = 17). At the end of the study follow-up, 8 patients (2.0%) remained active on the transplant waiting list.
Factors Associated With Dropout
In the univariate analysis, the following covariates were predictive of dropout due to tumor progression or death according to the CR analysis: multiple HCC tumors or a solitary tumor of 3.1 to 5 cm (versus a solitary tumor of 2-3 cm), a lack of a complete response to the first LRT, and the calculated MELD score (Table 3). The 1-year cumulative incidence of dropout was 6.2% (95% CI = 3.0%-10.9%) for patients with a single tumor of 2 to 3 cm, 16.6% (95% CI = 9.4%-25.7%) for patients with 2 tumors, 30.5% (95% CI = 22.9%-38.4%) for patients with a single tumor of 3.1 to 5 cm, and 46.4% (95% CI = 28.2%-62.8%) for patients with 3 tumors (P < 0.001; Fig. 1A). When the patients were stratified by their responses to the first LRT according to the modified RECIST criteria, the 1-year cumulative incidence of dropout was 9.3% (95% CI = 5.9%-13.6%) for patients with a complete response, 19.2% (95% CI = 11.1%-29.0%) for patients with a partial response, 39.5% (95% CI = 27.2%-51.1%) for patients with stable disease, and 85.0% (95% CI = 31.8%-97.7%) for patients with progressive disease (P < 0.001; Fig. 1B). The AFP levels at listing and after the first LRT both as continuous variables and at all cutoffs measured (including >1000, >500, >300, >100, and >20 ng/mL) were also predictive of dropout in the univariate analysis (Table 3). The 1-year cumulative incidence of dropout according to the AFP level after the first LRT was 12.7% (95% CI = 8.6%-17.6%) for patients with an AFP level ≤ 20 ng/mL, 20.7% (95% CI = 11.7%-31.4%) for patients with an AFP level of 21 to 100 ng/mL, 24.4% (95% CI = 9.9%-42.3%) for patients with an AFP level of 101 to 500 ng/mL, and 59.5% (95% CI = 35.3%-77.2%) for patients with an AFP level > 500 ng/mL (P < 0.001; Fig. 1C). Age, sex, liver disease etiology, Child cirrhosis class, and first LRT type were not significant predictors of dropout due to tumor progression or death (Table 3).
|Predictor||Univariate SHR (95% CI)||P Value|
|Patient characteristics at listing|
|Age (per year)||0.98 (0.96–1.01)||0.21|
|Female sex||0.93 (0.57–1.52)||0.78|
|Liver disease etiology (versus hepatitis C virus)|
|Hepatitis B virus||0.73 (0.44–1.21)||0.22|
|MELD score (per point)||1.04 (1.01–1.09)||0.049|
|Child C versus Child A||1.65 (0.87–3.12)||0.13|
|Child B versus Child A||1.20 (0.77–1.86)||0.42|
|Tumor characteristics at listing|
|3 lesions versus 1 lesion, 2–3 cm||7.76 (3.63–16.58)||<0.001|
|1 lesion, 3.1–5 cm, versus 1 lesion, 2–3 cm||4.31 (2.32–8.01)||<0.001|
|2 lesions versus 1 lesion, 2–3 cm||2.85 (1.42–5.72)||0.003|
|AFP level (ln)||1.21 (1.11–1.32)||<0.001|
|>1000 versus ≤1000 ng/mL||2.59 (1.42–4.72)||0.002|
|>500 versus ≤500 ng/mL||3.42 (2.03–5.74)||<0.001|
|>300 versus ≤300 ng/mL||2.35 (1.43–3.86)||0.001|
|>100 versus ≤100 ng/mL||2.01 (1.31–3.09)||0.001|
|>20 versus ≤20 ng/mL||1.91 (1.26–2.88)||0.002|
|Type of first LRT (versus TACE)|
|Radiographic and AFP response to first LRT|
|Partial response versus complete response||1.86 (0.98–3.54)||0.06|
|Stable disease versus complete response||5.75 (3.42–9.66)||<0.001|
|Progressive disease versus complete response||25.62 (12.03–54.54)||<0.001|
|AFP level after first LRT (ln)||1.29 (1.16–1.44)||<0.001|
|AFP level after first LRT|
|>1000 versus ≤1000 ng/mL||4.05 (1.82–9.00)||0.001|
|>500 versus ≤500 ng/mL||4.93 (2.70–8.98)||<0.001|
|>300 versus ≤300 ng/mL||3.28 (1.79–6.01)||<0.001|
|>100 versus ≤100 ng/mL||2.78 (1.64–4.72)||<0.001|
|>20 versus ≤20 ng/mL||2.16 (1.35–3.46)||0.001|
In the CR multivariate analysis (Table 4), the presence of 3 tumors [subhazard ratio (SHR) = 8.68, 95% CI = 3.25-23.19], the presence of 2 tumors (SHR = 5.03, 95% CI = 2.10-12.04), and the presence of a solitary tumor of 3.1 to 5 cm (SHR = 5.10, 95% CI = 2.28-11.41) were predictive of dropout due to tumor progression or death, as was a lack of a complete response to the first LRT (SHR = 3.08, 95% CI = 1.78-5.35). An AFP level > 20 ng/mL after the first LRT (SHR = 1.87, 95% CI = 1.13-3.10, P = 0.02) and all other AFP cutoffs in the univariate analysis were also significant predictors of dropout in the multivariate analysis. The MELD score was a predictor of dropout in the univariate analysis but not in the multivariate analysis.
|Predictor||Entire Cohort||Cohort After the Exclusion of Patients With Hepatitis B|
|Multivariate SHR (95% CI)||P Value||Multivariate SHR (95% CI)||P Value|
|3 lesions versus 1 lesion, 2–3 cm||8.68 (3.25–23.19)||<0.001||5.81 (3.25–23.19)||0.001|
|1 lesion, 3.1–5 cm, versus 1 lesion, 2–3 cm||5.10 (2.28–11.41)||<0.001||3.71 (1.64–8.40)||0.002|
|2 lesions versus 1 lesion, 2–3 cm||5.03 (2.10–12.04)||<0.001||3.06 (1.17–7.98)||0.02|
|Lack of a complete response to first LRT||3.08 (1.78–5.35)||<0.001||2.69 (1.44–5.02)||0.002|
|AFP level after first LRT (>20 versus ≤20 ng/mL)||1.87 (1.13–3.10)||0.02||2.33 (1.31–4.13)||0.004|
Subgroup With a Very Low Risk of Dropout
The 3 variables significantly associated with a lower risk of wait-list dropout due to tumor progression or death included a single, 2- to 3-cm lesion on presentation; a complete response to the first LRT; and a low AFP level after the first LRT. In order to identify a group of HCC patients with the lowest risk of wait-list dropout, we used the lowest AFP cutoff significantly associated with dropout (20 ng/mL). Complete data for each of these 3 variables was available for 317 of the 398 patients (79.6%). Among these 317 patients, a subgroup of 63 patients (19.9%) met the following criteria: 1 tumor of 2 to 3 cm, a complete response to the first LRT, and an AFP level ≤ 20 ng/mL after the first LRT. This subgroup had estimated 1- and 2-year probabilities of dropout or death without LT of 1.3% and 1.6%, respectively, whereas the probabilities were 21.6% and 26.5% for all other patients (P = 0.004; Fig. 2).
We also performed a sensitivity analysis and took into consideration the high prevalence of hepatitis B in our cohort, which is likely different from the prevalence in many other transplant centers across the country. When we excluded patients with hepatitis B and used the same criteria for a low risk of dropout, the 1- and 2-year probabilities of dropout were 2.2% and 2.5%, respectively, for the low-risk group and 22.5% and 27.0%, respectively, for all other patients (P = 0.02). Additionally, each component of the low-risk group (1 lesion of 2-3 cm, a complete response to the first LRT, and an AFP level ≤ 20 ng/mL) remained significantly associated with reduced wait-list dropout in the multivariate analysis after the exclusion of all hepatitis B patients (Table 4).
Clinical characteristics, the number of LRTs received, and histopathological characteristics were compared between the subgroup with a low risk of dropout and all other patients (Table 5). Although the listing MELD scores were similar, the low-risk group had a lower percentage of Child C patients (1.6% versus 9.8%, P = 0.03). Patients at low risk of wait-list dropout were significantly more likely to receive a single LRT (82.5% versus 44.9%, P < 0.001). LT was performed for 57 of the 63 patients (90.5%) in the low-risk group and for 167 of the 254 patients (65.7%) in the high-risk group. Patients in the low-risk group were more likely to have complete tumor necrosis on explant (60.7% versus 34.0%, P < 0.001) and were less likely to have an explant tumor burden outside the T2 criteria (3.5% versus 21.6%, P = 0.04). The rates of macrovascular or microvascular invasion in the explant did not differ significantly between the 2 groups.
|Characteristic||Low-Risk Group||All Others||P Value|
|Listing MELD score ≥ 15 [n (%)]||10 (15.9)||45 (17.7)||0.73|
|CPT score [n (%)]|
|A||34 (54.0)||132 (52.0)||0.66|
|B||28 (44.4)||97 (38.2)||0.36|
|C||1 (1.6)||25 (9.8)||0.03|
|Number of LRTs [n (%)]|
|1||52 (82.5)||114 (44.9)||<0.001|
|2||7 (11.1)||80 (31.5)||0.001|
|3||2 (3.2)||37 (14.6)||0.01|
|>3||2 (3.2)||23 (9.1)||0.12|
|Pathological stage [n (%)]a|
|No residual tumor||35 (61.4)||59 (35.3)||0.001|
|T1||3 (5.3)||15 (9.0)||0.02|
|T2||17 (29.8)||57 (34.1)||0.22|
|>T2||2 (3.5)||36 (21.6)||0.04|
|Vascular invasion [n (%)]a||2 (3.5)||8 (4.8)||0.69|
|Histological grade [n (%)]b|
|Completely necrotic||34 (60.7)||55 (34.0)||<0.001|
|Well differentiated||6 (10.7)||45 (27.8)||0.009|
|Moderately differentiated||12 (21.4)||58 (35.8)||0.047|
|Poorly differentiated||4 (7.1)||4 (2.5)||0.11|
The demand for liver allografts far exceeds the supply in the United States, in part because of the rising incidence of HCC and the growing numbers of patients with HCC who may benefit from LT. Since the MELD allocation scheme for HCC was implemented in 2002, the proportion of patients receiving a priority listing with an HCC MELD exception has increased from 10.5% in 2002 to 15.5% in 2008. A pressing problem facing the transplant community is the growing body of evidence showing that patients with HCC are given an unfair advantage in organ allocation in comparison with non-HCC patients listed for LT according to their calculated MELD scores.[9-12] Using UNOS data from 2005 to 2008, Washburn et al. analyzed the rates of wait-list dropout by CR statistics and found significantly higher rates of wait-list dropout for non-HCC patients versus HCC patients. This observation was reproducible across all regions, although the disparities were relatively small in 2 of the 11 regions. Goldberg et al. analyzed the UNOS database from 2005 to 2009 and similarly demonstrated that the odds of wait-list removal due to death or clinical deterioration were significantly lower for HCC patients versus non-HCC patients. Furthermore, the observed differences between the 2 groups in wait-list removal increased steeply as the analysis progressed from the lower MELD strata to the higher strata. At a MELD score of 22 for LT candidates with HCC, 4.6% were removed from the waiting list within 90 days, whereas the corresponding wait-list removal rate was 11% for non-HCC candidates with MELD scores of 21 to 23. The wait-list removal rate within 90 days was only 3% for HCC patients with 28 MELD exception points, whereas the rate was 23.6% for non-HCC patients with MELD scores of 27 to 29.
Although it is generally recognized that changes are needed to address the disparity in the rates of wait-list dropout for HCC and non-HCC patients, there is no easy solution to this problem. Reducing the initial priority score for patients with HCC at the national level does not account for the significant regional variations in wait-list times as well as wait-list removal rates. This strategy of lowering the MELD score uniformly across all regions may in fact place HCC patients at a disadvantage in certain regions. A different approach of giving a higher listing priority to those at greater risk of dropout from the waiting list has been advocated.[9, 13, 18] Freeman et al. proposed the HCC MELD equation, which is a continuous score based on the calculated MELD score at listing, the AFP level, and the maximum tumor size: the higher the score is, the greater the risk is for wait-list dropout. This score has been proposed in a scheme that gives the highest organ allocation priority to those with the greatest risk of wait-list dropout as predicted by the HCC MELD equation. Toso et al. analyzed wait-list dropout in the Scientific Registry of Transplant Recipients and proposed a common waiting list including both HCC and non-HCC patients, who were assigned a continuous dropout equivalent MELD score based on their age, calculated MELD score, tumor size and number, AFP level, and liver disease etiology. However, proposals to give priority to patients with higher AFP levels and larger tumors have also raised concerns about the selection of tumors with a worse biology and inferior posttransplant outcomes. A high AFP level has been shown in a plethora of studies to be predictive of a worse prognosis after LT, especially when the level exceeds 400 or 1000 ng/mL. There is also good correlation between a larger tumor diameter and a greater likelihood of microvascular invasion. In fact, a recent study by Cucchetti et al. using the HCC MELD equation has shown that the higher the probability of wait-list dropout is as predicted by the HCC MELD equation, the lower the survival will be after LT.
In the present study, we tested the hypothesis that there is a subgroup of patients with a very low dropout risk who do not require the same listing priority. Using CR regression, we found that the subgroup meeting all 3 criteria—a complete response after the first LRT, an AFP level ≤ 20 ng/mL after the first LRT, and a single lesion of 2 to 3 cm—had a very low probability of wait-list dropout (only 1.3% at 1 year and 1.6% at 2 years versus 21.6% and 26.5%, respectively, for all other patients). This group with a very low dropout risk accounted for approximately 20% of our cohort. Reducing the listing priority for this group would decrease the burden for HCC patients receiving MELD exceptions who truly need timely LT and allow non-HCC patients to have greater access to LT. This may potentially reduce the inequities in dropout rates for non-HCC and HCC patients.
In contrast to several other published reports,[7, 9, 18] the calculated MELD score at the time of listing was predictive of wait-list dropout only in the univariate analysis and not in the multivariate analysis of our study. Washburn et al. found a 9% increase in wait-list dropout with each additional MELD point at listing. The difference may be explained by the high proportion of patients with well-compensated cirrhosis in our cohort, for which the median MELD score was only 11, and only 22.6% of our patients had a MELD score of 15 or higher. Only 3.5% of the patients in our cohort did not receive LRT, and among those patients, only 28.6% had very advanced decompensated liver disease. Patients who are really too sick to receive any LRT likely have calculated MELD scores exceeding the HCC MELD exception and are allocated organs on the basis of their true MELD score. It should also be pointed out that the benefit of LRT in reducing the rate of wait-list dropout has not been confirmed because of the lack of randomized controlled trials, but such a study is not likely to be feasible. LRT is recommended, however, when the wait-list time is expected to be at least 6 months. Our observation that a complete response to LRT correlates with a lower probability of wait-list dropout is consistent with the findings in a previous study by Cucchetti et al. Their study included 315 patients: 30% had HCC outside the Milan criteria before LT, and 93% had received LRT. The median wait-list time was 10 months, and the 1-year probability of dropout was 19.9%. They found a complete response to LRT (45.7% of their patients) to be the only significant predictor of a low risk of dropout in a multivariate CR analysis.
It may be argued that using the response to the first LRT as a means of determining listing priority is not applicable to regions with short wait-list times, in which the benefits of LRT have not been established. Kadry et al. reported regional variations in the use of LRT before LT for HCC. Regions with a median waiting time > 6 months performed LRT more and had significantly higher wait-list dropout rates. Nevertheless, the differences in the rates of LRT were not as dramatic as one would expect from the disparities in wait-list times. In 8 of the 11 regions in which more than 80% of patients received LT within 6 months, the proportion of patients receiving LRT before LT ranged from 36% to 53%, whereas the range was 50% to 54% for the other 3 regions with longer wait-list times. The need to observe the tumor response after the first LRT in our current proposal also incorporates the ablate-and-wait principle, which is designed to prevent transplantation for patients whose tumors have a poor biology and progress rapidly despite LRT and who also do poorly after LT. Our cohort had a high prevalence of hepatitis B and a very low prevalence of fatty liver disease, so it may not fit the usual demographics of most other transplant centers across the country. By excluding hepatitis B patients, we performed a sensitivity analysis, and we found similar results, with a significantly lower risk of dropout for patients meeting certain criteria (a complete response after the first LRT, an AFP level ≤ 20 ng/mL after the first LRT, and a single lesion of 2-3 cm) versus patients not meeting these criteria.
Our study has several limitations, most notably the retrospective study design and the lack of complete information on radiographic and biochemical (AFP) responses to the first LRT in 20% of patients, in part because we included only data obtained within 1 month before and after the first LRT. There are also several strengths of our study. Our center is situated in region 5, which has the nation's longest wait-list time and the highest 6- and 12-month dropout rates: this provides a rather unique position for accurately assessing factors predictive of wait-list dropout. The use of LRT in almost all of the patients in our cohort also allows us to evaluate the impact of the response to LRT on the risk of wait-list dropout, whereas this information was not readily available in previous analyses of large national databases.[7, 9, 18] Similarly to Washburn et al., we used the CR method to evaluate the rates and predictors of dropout from the waiting list rather than the Kaplan-Meier method, which has been used in a number of previous studies.[4, 13, 19] The CR method more accurately estimates the cumulative incidence than the Kaplan-Meier method: in the latter, censored and transplant patients are treated in the same way (both are censored), and the cumulative incidence is overestimated.
In summary, the present study suggests that a combination of tumor characteristics and a complete response to the first LRT defines a subgroup with a very low wait-list dropout risk of 1.3% within 1 year and 1.6% within 2 years. The vast majority of these patients will not derive an immediate benefit from LT and should not receive the same listing priority as other patients with T2 HCC. Under the current climate in which the demand for organs far exceeds the supply and HCC patients have an advantage in the organ allocation scheme, our results may have important implications for the organ allocation policy for HCC.
- 3UNOS/OPTN Liver Disease Severity Score, UNOS/OPTN Liver and Intestine, and UNOS/OPTN Pediatric Transplantation Committees. The new liver allocation system: moving towards evidence-based transplantation policy. Liver Transpl 2002;8:851-858., , , , , , et al.; for
- 4A follow-up analysis of the pattern and predictors of dropout from the waiting list for liver transplantation in patients with hepatocellular carcinoma: implications for the current organ allocation policy. Liver Transpl 2003;9:684-692., , , , , , et al.
- 25A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999;94:496-509., .
- 27Impact of geographic disparity on liver allocation for hepatocellular cancer in the United States. J Hepatol 2012;56:618-625., , , , , .